T E C H N I C A L N O T E

In Vivo Preclinical ImagingAutomated Bone Mineral Density (BMD) Calculations Using the Quantum GX microCT Imaging System and AccuCT Advanced Imaging Software

The QuantumGX x-ray microCT preclinical imaging system offers high resolution combined with high speed imaging capabilities to enable research over a wide variety of applications including oncology, cardiovascular and pulmonary disease, metabolic syndromes and orthopedics. Skeletal imaging using microCT is common, given that mineralized bones possess excellent x-ray attenuation properties, thus making them easily distinguishable from soft tissue and air. However, analysis of skeletal microCT images tends to be time-consuming and subjective due to the potential need for manual detection and separation of bones. The AccuCT™ imaging software provides advanced, yet user-friendly, automated workflows to robustly detect, separate and report bone-related measurements such as ASBMR morphometry parameters, volume for bone growth or loss studies, and bone mineral density (BMD).

Certain skeletal research areas, such as osteoporosis and osteoarthritis, utilize BMD as a metric to report on disease state. The most accurate measurements of BMD using a microCT

scanner require the use of a BMD calibration phantom, which is commonly comprised of rods of varying hydroxyapatite (HA) densities housed in a resin cylinder. The microCT scans of these phantoms allow researchers to calibrate their experimental scans from Hounsfield units or grayscale values into the BMD unit of mg HA/cm3. AccuCT automates and streamlines the typical calibration process of manual rod segmentation, linear regression and image algebra.

This technical note provides an overview on the selection and microCT scanning of BMD phantoms as well as the use of AccuCT to generate BMD measurements from QuantumGX microCT scans. AccuCT provides automated workflows to segment BMD phantoms and generate calibration plots, and associate experimental scans with their appropriate BMD phantom calibration plot. Single bone or whole scan BMD measurement automated workflows are also provided with AccuCT and highlighted in this technical note.

Authors:

Julie Czupryna, Ph.D.

Jeff Meganck, Ph.D.

Ali Behrooz, Ph.D.

PerkinElmer, Inc. Hopkinton, MA

2

Selecting and Scanning the Proper BMD Phantom

The AccuCT BMD calibration workflow automatically segments inserts within QRM-MicroCT-HA phantoms. These inserts contain various densities of hydroxyapatite (HA) and are housed in a cylindrical resin base structure, as shown in Figure 1.

the 25 mm FOV does not include the entire phantom (Figure 2e) and therefore, a smaller diameter BMD phantom should be chosen for that scan FOV.

A unique feature of the QuantumGX microCT is the ability to perform sub-volume reconstructions on acquired scans. Sub-volume resconstructions offer researchers flexibility: researchers can acquire a larger scan FOV and then perform a subsequent reconstruction on a user-defined region of interest (ROI), resulting in a volume with improved spatial resolution. This process eliminates the need for multiple scans of a single subject, ultimately reducing longitudinal radiation dose. It is important to note that if BMD analyses will be performed on a sub-volume reconstruction, then the same type of sub-volume reconstruction should be applied to the BMD phantom. As mentioned above, the entire BMD phantom still needs to be included in the newly reconstructed volume. Finally, researchers should verify that the grayscale/HU values of the sub-volume reconstructions match those of the original scans.

Once the appropriately sized BMD phantom is chosen, it should be scanned using not only the same FOV, but also the same voltage, current and scan time as the intended experimental samples. Scanning the phantom once per imaging session/day can provide a valuable quality control metric, however we recommend that image calibrations for BMD data analysis be determined by local protocol for daily, weekly, monthly or study-based phantom scanning. AccuCT allows for fast, streamlined calibration of multiple BMD phantoms within a single workflow, if necessary.

Figure 2. 32 mm QRM-MicroCT-HA phantom at various QuantumGX scan FOVs. (a) 72 mm, (b) 60 mm, (c) 45 mm, (d) 36 mm and (e) 25 mm. Scanning the 32 mm BMD phantom using the 25 mm FOV is not recommended.

a c eb d

Automated BMD Phantom Calibration With AccuCT

The AccuCT software offers an automated workflow for creating a BMD calibration plot from a BMD phantom scan. The microCT scan of the BMD phantom is loaded into the Calibration Data tab of the AccuCT study (Figure 3a) and the phantom is visualized in both 2D and 3D (Figure 3b). During scan selection, the size of the phantom is also selected. The HA inserts are automatically detected using the Segment BMD Phantom step (Figure 3c). To do this, a multi-thresholding algorithm is used to detect the voxels belonging to HA inserts based on the histogram of the voxel intensity values of the BMD phantom scan. After image post-processing which includes morphological filtering and geometrical mapping, the individual HA inserts are segmented and labeled separately using connected component analysis as illustrated in Figure 3.

Figure 3. AccuCT BMD Calibration workflow. (a) BMD phantom scan and calibration workflow list. (b) 2D and 3D visualization of the segmented phantom. (c) Calibration workflow. (d) Known insert HA density information. Inset: calibration plot that can be seen in the Plots tab.

a

c

bd

Figure 1. (L) The 32 mm QRM-MicroCT-HA phantom. (M) A schematic of the 5 HA inserts within the phantom. (R) A 2D slice from a representative microCT scan of the 32 mm phantom.

In order for the AccuCT workflow to create a calibration curve based on the phantom inserts, it is important to ensure that all five inserts and the entire diameter of base resin are fully included in the microCT scan field of view (FOV). Additionally, the phantom should be scanned using the same FOV as the intended experimental subjects. Therefore, the size of the chosen phantom may vary depending on the scan field size. A range of phantom sizes is available from 32 mm down to 4.5 mm to suit most microCT imaging needs. Figure 2 highlights various scan FOVs using the 32 mm phantom in the QuantumGX. It can be seen that the 72 mm, 60 mm, 45 mm and 36 mm FOVs include the entire diameter of the phantom. However,

For a complete listing of our global offices, visit www.perkinelmer.com/ContactUs

Copyright ©2017, PerkinElmer, Inc. All rights reserved. PerkinElmer® is a registered trademark of PerkinElmer, Inc. All other trademarks are the property of their respective owners. 013463_01 PKI

PerkinElmer, Inc. 940 Winter Street Waltham, MA 02451 USA P: (800) 762-4000 or (+1) 203-925-4602www.perkinelmer.com

It must be noted that this automated approach makes no assumptions regarding the spatial location or orientation of the BMD phantom during the scan. As a result, the performance of the automated approach is invariant with respect to the BMD phantom placement or positioning in the scanner instrument. After segmentation, the mean values of the rod densities are extracted from each labeled HA insert. Then the user verifies the values of the rod densities (Figure 3d) and AccuCT generates a calibration plot (Figure 3 inset). This is done by performing a linear regression fit on the density data with pre-known BMD values of the HA inserts used as y-axis data and mean values of the rod densities used as x-axis data. The result of the regression fit provides the slope and y-intercept associated with the BMD calibration. These calibration factors can then be used on any scan with the same parameters to perform BMD calibration. Multiple BMD phantom scans can be calibrated within the same AccuCT study, if necessary.

Associating MicroCT Scans With BMD Calibration Factors

Once BMD phantom calibration is complete, experimental scans can be associated with the calibration in order to provide BMD measurements. After experimental scans are loaded into the Animals & Data tab of the AccuCT study, the association begins by right clicking the experimental scan of interest and selecting the first choice of Associate BMD calibration (Figure 4a). The appropriate calibration scan is selected (Figure 4b) and the AccuCT software automatically converts the experimental scan to the unit of mg HA/cm3. Experimental scans that have been associated with a BMD calibration will be annotated with a green ‘c’ (Figure 4c).

Obtaining BMD Measurements Using AccuCT

Following BMD calibration association, AccuCT provides automated workflows for single bone or whole scan BMD measurements. The single bone BMD workflow is shown in Figure 5 illustrating bone detection and separation steps in order to select the bone of interest for BMD analysis. The bone detection and separation steps contain settings that can be customized such as boundary hardness and minimum size (detection) and sensitivity (separation). In the event of under- or overseparation, there are steps to segment more or join segments, respectively. Once the appropriate separation is achieved, the bone of interest is selected and the reported BMD measurements can be seen in the inset of Figure 5.

Figure 4. Association of a BMD calibration. (a) Right click experimental scan and choose Associate BMD calibration (b) Choose appropriate calibration (c) Green ‘c’ appears next to experimental scan to indicate association.

a

c

b

Figure 5. AccuCT Single Bone BMD Workflow semi-Automated workflow steps are found in the bottom of the AccuCT software window. The bone selected for BMD analysis is indicated in red. BMD measurement results are shown in the inset.

Summary

This technical note provides researchers with an overview of BMD phantom selection and microCT scanning using the QuantumGX. It also highlights the use of the AccuCT imaging software platform to automatically calibrate experimental scans for BMD analyses using the BMD phantom scans. AccuCT provides automated workflows for BMD measurements following calibration, on either single bones or entire scan volumes. User-definable settings allow for model-specific customization of analyses within the automated workflows, with the overall goal of consistency across analyses, independent of the AccuCT user.